swarm-research

Pass

Audited by Gen Agent Trust Hub on Jul 24, 2026

Risk Level: SAFEPROMPT_INJECTIONNO_CODE
Full Analysis
  • [PROMPT_INJECTION]: The skill presents an attack surface for indirect prompt injection by relaying untrusted user input to multiple sub-agents within a structured workflow.
  • Ingestion points: The research topic or question provided as a primary argument to the skill.
  • Boundary markers: There are no explicit markers or delimiters defined to isolate the user input from the system instructions used to prompt worker agents.
  • Capability inventory: The orchestration workflow involves reading and writing files to the scratchpad/ and artifacts/ directories, as well as managing tasks via internal tools.
  • Sanitization: No sanitization or validation of the input research topic is mentioned in the instructions.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 24, 2026, 10:41 PM
Security Audit — agent-trust-hub — swarm-research